Differential protein expression analysis

Question

[Ribo]

Hi,

I have proteomic data with a few replicates per cell type, and I would like to find the significant differential expressed proteins between cell types. Each individual protein raw intensity value was divided by the total intensity detected for each technical replicate and multiplied by one million.

1. What statistical test should I use?
2. Should I log-transform the data before testing?
3. Is there any recommended R package for such analysis?

Thank you all in advance

[dsull]

Yes, you need to log-transform.

I've had a good experience using limma for proteomics analyses (limma uses a moderated t-test).

[Ribo]

Thank you!

Is there a recommended tutorial for proteomics analysis?

[dsull]

I don't have a tutorial but see this paper I co-authored:

https://pubs.acs.org/doi/abs/10.1021/acs.jproteome.0c00666

All of the upstream processing: intensity / normalization / spectral library search etc. were done via Scaffold DIA, but I have code (written in R) included in the GitHub repo linked to in that paper to do differential expression.

On a separate note, an unfortunate thing about proteomics (compared to sequencing data) is that most processing is done using proprietary software (Byonic, Scaffold, MaxQuant, etc.).